Are Naturalized Voters Driving the California Latino Electorate? Measuring the Effect of IRCA Citizens on Latino Voting n

Transcription

1 Are Naturalized Voters Driving the California Latino Electorate? Measuring the Effect of IRCA Citizens on Latino Voting n Matt A. Barreto, University of Washington Ricardo Ramírez, University of Southern California Nathan D. Woods, Welch Consulting Objectives. The 1990s witnessed the growth and maturation of the Latino electorate in California and many scholars have posited as to the reasons. One argument is that naturalizations by way of the Immigration Reform and Control Act (IRCA) drove the increases in Latino participation. In this article we investigate the extent to which this is the case. Methods. Using unpublished INS data, we offer the first empirical test of the IRCA theory by examining Latino IRCA petitions by zipcode to determine whether or not IRCA legalizations and subsequent naturalizations were the force behind increased Latino turnout, and the overall growth of the Latino vote. We merge IRCA data with Registrar of Voter data to examine real growth in the Latino vote at the zipcode level from in southern California. Results. Although Latino voting grew substantially, we find that IRCA naturalizations did not spur the increases in Latino voting in the 1990s as some have expected. Instead, demographic and mobilization variables explain why the Latino vote grew between 1996 and Conclusions. As Congress debates new proposals to legalize the millions of undocumented immigrants living and working in this country, many will inevitably ask what impact their citizenship will have on the electorate. This study sheds some light on the relationship between amnesty programs, citizenship, and voting among Latinos. The 1990s witnessed the growth and maturation of the Latino electorate in California and many scholars have posited as to the reasons. Population n Author names are presented in alphabetical order. For questions or access to the data set, please direct correspondence to Ricardo Ramírez hramirez1@usc.edui, who will make the data and coding instructions available to any interested parties. The authors acknowledge the helpful comments from Adrian Pantoja, Gary M. Segura, and Yishaiya Abosch on an earlier draft of this article. In addition, the authors are indebted to Harry Pachon and the Tomás Rivera Policy Institute for making available validated data on Latino voting in California and to Rosalind Gold of NALEO for sharing INS data on IRCA petitions. Any errors are the authors alone. SOCIAL SCIENCE QUARTERLY, Volume 86, Number 4, December 2005 r2005 by the Southwestern Social Science Association

2 Effect of IRCA Citizens on Latino Voting 793 growth, the increase in the pool of citizen voting-age population, voter registration mobilization efforts, and a reaction to race-targeting ballot initiatives have all been heralded for awakening the sleeping giant in California politics. The Latino share of the population increased from 25.8 percent in 1990 to 32.4 percent in During this same period, there was a tremendous increase in the number of Latino immigrants naturalizing, both numerically and as a percent of all California naturalizations. In 1990, immigrants born in Latin America comprised only 9.19 percent of those naturalizing. Ten years later, the figure had soared to 41.6 percent. 1 Latino civic organizations have contributed to this increase in Latino naturalized citizens and are also partly responsible for the increase in the number of Latinos registered to vote and those actually voting. The presence of Latinos in electoral politics has grown since 1992 when Latinos accounted for only 7 percent of the statewide electorate. By 2000, Latinos comprised 14 percent of the electorate, an increase of over 1 million registered voters. Clearly, Latino population growth, levels of naturalization, and mobilization efforts have helped shape the role of Latinos in California politics, but the specific contribution of each factor to increasing Latino political participation in California is still being debated. What makes it difficult to disentangle the net effects of the above-mentioned variables is the extent to which the changes in the state s political landscape were impacted by national forces, California-specific factors, or the multiplicative effects of the interaction between national and state factors (Ramírez, 2002; Fraga and Guerra, 1996; Pachon, 1998; Pantoja, Ramírez, and Segura, 2001; Segura, Falcon, and Pachon, 1997; Pantoja and Segura, 2003; Barreto, Segura, and Woods, 2004). In this article, we look specifically at one such national force, the Immigration Reform and Control Act of 1986 (IRCA), and consider the extent to which it was responsible for changes in the increased Latino presence at the polls, as suggested by Valle and Torres (2003). Although our article focuses on a historic amnesty proposal, it may hold implications for the future, given that both the Republican and Democratic Parties are currently discussing legalization and/or amnesty proposals for the estimated 7.5 million undocumented Latin American immigrants living in the United States (Lindlaw, 2004). Using a combination of unpublished INS administrative data on IRCA petitions, Census data, and Registrar of Voters data, we offer the first empirical test to determine the extent to which there is a link between the concentration of Latinos who obtained legal residence through IRCA and the increased pool of the Latino citizen voting-age population. Additionally, by using Registrar of Voter records, we can identify the date of registration and validated vote entries for all registrants in the analysis. This will allow us 1 This figure still represents a drop from a high of 50 percent in FY 1996 (Statistical Yearbook of the INS, 1996).

3 794 Social Science Quarterly to consider the effects of concentration of IRCA residents and nativity alongside controls for first registering to vote in the turbulent era. In short, we examine the effects of the number of IRCA residents in a zipcode on two dependent variables: (1) the voter turnout level of Latinos in the 1996 and 2000 general elections and (2) the growth in Latino voters between 1996 and The rest of this article is divided into five sections. In the first section, we review the primary explanations for the growth in the Latino electorate in the United States and California. In the second section, we give an overview of legalization and naturalization trends since 1990 in California and the United States and we discuss their potential to shape the Latino electorate. In the third section, we describe the data and methods used. The fourth section presents the results and the final section concludes with a discussion of the consequence of the findings. Leading Explanations for the Growth in the Latino Electorate There are four primary explanations given for the growth in the national Latino electorate. Three of these apply to Latinos throughout the United States, while one is specific to California. Some observers question the media s depiction of Latinos participating at higher rates currently than in previous elections, and argue that these increases are driven by population growth, rather than an increase in voter turnout (de la Garza, Haynes, and Ryu, 2002; Shaw, de la Garza, and Lee, 2000). In fact, in the decade of the 1990s there was a dramatic increase in the pool of U.S.-born Latinos of voting age. In 1990 there were 3 million Latinos age 18 to 24. By 2000, there were 4.7 million Latinos age 18 to 24. Although some of the increase in Latinos of voting age can be attributed to unauthorized immigrants, it is safe to say that most of these Latino youth are U.S. citizens. The second explanation is the significant increase in Latino naturalized citizens in the mid 1990s (Pachon, 1999). In particular, there are three often-cited causes of the increased number of Latino naturalized citizens: IRCA immigrants eligible to naturalize, the INS Green Card replacement program, and welfare reform. First, starting in 1993, the first cohort of immigrants who were granted legal permanent residency under the Immigration Reform and Control Act of 1986 became eligible for naturalization. 2 According to some, the IRCA effect was a leading cause of Latino vote growth. Valle and Torres argue that California s Latino immigrants took advantage of a period of immigration amnesty, became citizens, and then 2 Among the most basic naturalization requirements is the need to have resided continuously as a lawful permanent resident in the U.S. for at least 5 years prior to filing with no single absence from the United States of more than one year (Bureau of Citizenship and Immigration Services).

4 Effect of IRCA Citizens on Latino Voting 795 voted in record numbers (2003:389). Specifically, they note that the 1996 election results show that newly enfranchised Latino immigrants voted at a rate exceeding that of the state s voters as a whole (Valle and Torres, 2003:389); however, their research methodology focused on interviews with activists, not data analysis of official vote records or IRCA petition data. Although IRCA expanded the pool of potential Latino citizens, it does not explain why Latinos would behave differently than they had in the past, where the time lag between legal residency and naturalization has been particularly high. 3 The INS Green Card replacement program and the resulting cost-benefit analysis by immigrants is also cited as a cause for the increase in number of naturalizations. Specifically, legal permanent residents (LPRs) are presumed to have made the decision to pay the one-time fee of $95 for the application to naturalize as opposed to the mandatory $75 fee to renew the Alien Service Receipt Card, which a 1996 Immigration and Naturalization rule states must be updated every 10 years. 4 In an effort to minimize fraud, the 1996 INS rule required that all permanent residents holding the old Form I-151, Alien Registration Receipt Card, which was issued before 1979, must apply in person as soon as possible for a new card, Form I-551, Alien Registration Receipt Card, to demonstrate satisfactory proof of permanent resident status and work eligibility in the United States. A third cited cause of increased naturalization rates contends that the decision to naturalize was a strategic political act in reaction to the enactment of the Personal Responsibility and Work Opportunity Reconciliation Act of This Act had the effect of denying welfare benefits to legal immigrants. The third explanation for the growth in the Latino electorate is the effort to mobilize voters by Latino organizations targeting Latinos, but in many cases focusing on the above groups of newly eligible citizen voting-age Latinos because that is where the growth in the electorate is most likely to happen. 5 The National Association of Latino Elected and Appointed Officials (NALEO) and Southwest Voter Registration and Education Project (SVREP) have conducted the largest efforts to increase the potential for Latinos to have a voice in local, state, and national politics. During the 3 For example, the median number of years of residence between the date of legal permanent residence and date of naturalization for immigrants from North America tends to be about 11 years, but only seven years for European immigrants. In this regard, though, note that the Bureau of Citizenship and Immigration Services considers the Caribbean and Central America in its totals for North America. 4 Previously, this fee applied only to those who had to replace lost or stolen cards. From this point on, legal residents will incur the card replacement cost every 10 years. The key point is simply that, one way or another, Latino legal residents had to pay a fee. Beyond the other benefits incurred by citizenship, a simple cost-benefit analysis would suggest a preference for a one-time fee versus recurring fees and time spent on the replacement process. 5 The other potential source is of the segment of the population that is already eligible to vote but has remained inactive.

5 796 Social Science Quarterly 1990s, NALEO was instrumental in the efforts of Latinos to naturalize by assisting them with naturalization petitions. On the other hand, through its Su Voto Es Su Voz (Your Vote is Your Voice) campaign, SVREP has largely focused on registering Latinos to vote. More recently, both organizations have expanded their get out the vote (GOTV) efforts more systematically with special efforts to target infrequent voters. 6 However, the systematic nature of such efforts is a more recent phenomenon. 7 Additional research by Shaw, de la Garza, and Lee (2000) and Nuño (2005) has found that Latin voters are often persuaded when the mobilization is done by other Latinos. Finally, it has been posited that a unique political context in California in the mid-1990s triggered the increased naturalization rates as well as the increased voter registration and turnout among Latinos in that state. Furthermore, it has been suggested that this political context in conjunction with the increased pool of IRCA legal permanent residents (LPRs) and U.S.- born Latino youth led to increased registration turnout rates. Proposition 187, the 1994 anti-immigrant state initiative, drew national attention and potentially served to mobilize the Latino electorate by prodding both indifferent and first-time voters to the polls. Accordingly, 1994 represents a watershed year for Latino politics. It is further posited that two additional race-targeting ballot initiatives (209 in 1996 and 227 in 1998) further served to mobilize Latino participation. Pantoja, Ramirez, and Segura (2001) find that Latinos who naturalized between 1992 and 1996 in California were more likely to vote than their newly naturalized counterparts in Florida and Texas, which had no corresponding salient and divisive state ballot measures in the same period. 8 Another effort attempts to determine the partisan consequences of divisive state ballot measures on electoral participation of Latinos and non-latinos between 1992 and 1998 in Los Angeles County (Barreto and Woods, 2005). Using Registrar of Voter data, they find that new Latino registered voters were less likely to register as Republicans, whereas the Democratic Party was not seriously impacted and third parties saw improved figures. The partisan implications among these newly mobilized Latinos who first registered between 1992 and 1998 in Los Angeles County (62.8 percent are registered Democrats, whereas only 10.3 percent were registered Republicans) make 6 For an analysis of the effectiveness of GOTV efforts by NALEO and SVREP, see Ramirez (2002) and Pantoja and Woods (1999). 7 SVREP conducts a Latino Academy to train interested individuals and organizations to conduct Latino voter registration and mobilization efforts. NALEO s Voces Del Pueblo program is intended to increase Latino turnout by specifically targeting low-propensity Latino registered voters through public service announcements and targeted mail and phone contact. 8 They make reference to these years because this was when the greatest increase in naturalization and Latino electoral participation occurred, and they attribute this growth primarily in California to the politicized climate instigated by Propositions 187 in 1994 and 209 in 1996.

6 Effect of IRCA Citizens on Latino Voting 797 the advantage gained by Democrats during the New Deal realignment pale in comparison. Additionally, when comparing turnout patterns among Latinos in California, others have found that those who first registered to vote at the height of the politicized context (between 1994 and 1996) consistently had higher rates of turnout between 1996 and 2000 than Latinos who first registered before or after this politicized period of time (pre 1994 or post 1996), and that within this unique registration cohort, Latino naturalized citizens had the highest rates of participation (Ramirez, 2002). In full, each of these factors may have some effect on the increase in Latino voting. In subsequent sections we discuss trends in IRCA-related naturalization in greater detail and develop hypotheses about its anticipated effects on voting. IRCA Legalization and Naturalization Trends California s political landscape changed definitively in the mid-1990s (Segura, Falcon, and Pachon, 1997). But how did the California context interact with national-level factors? Here, we focus on one national factor: IRCA legalizations. In 1986, Congress passed the Immigration Reform and Control Act, which granted residency to nearly 2.7 million undocumented residents, the majority of whom were from Latin America. There were two groups of immigrants eligible for legal permanent status under IRCA provisions. The first group included unauthorized immigrants who had been residing in the United States since before January 1, 1982 (Pre-1982s) and the second group included immigrants employed in seasonal agricultural work (SAWs) for a minimum of 90 days in the year prior to May Starting in 1992, the first wave of those who obtained legal resident status through the provisions of IRCA became eligible to apply for citizenship. Some observers have not only credited this new pool of eligible citizens for the surge in naturalizations in the mid-1990s but also for the significant growth in the Latino vote during this period. The passage of Proposition 187 is the most often-cited explanation given for why Latino IRCA residents are distinct from previous cohorts of Latino residents eligible to naturalize (characterized by their sluggishness between when they have been eligible to naturalize and when they actually undergo the process). However, beyond the rhetoric of identifying which segments of the Latino electorate were mobilized voters, it is not clear that Latino IRCA residents were a significant segment of the newly naturalized or of the burgeoning Latino electorate in the middle of the decade. Although Valle and Torres point out that 1.7 million immigration applicants all became eligible to apply for amnesty at the end of 1995, and by 1996, these applicants became citizens in increasing numbers (2003:389), there is no evidence that they were voting in large numbers.

7 798 Social Science Quarterly Why is it important to identify whether IRCA is driving the Latino electorate? The implications for Latino incorporation into American politics and society are negative should IRCA immigrants account for most of the growth in naturalization and electoral participation. If the significant increase in the pool of citizen voting-age population can largely be explained by the presence of IRCA immigrants, then increases in participation may not be sustained and the prospects for continued social and political integration of Latinos would be low. Independent of the political context, were IRCA legal permanent residents (LPRs) so unique that they flooded the Immigration and Naturalization Service offices by applying in droves the first year that they were eligible? The records indicate otherwise. In FY 1993 there was a surge in naturalizations of about 75,000 additional new citizens than in FY 1992, up to 314,681. However, only 881 of the naturalized citizens were IRCA immigrants (Statistical Yearbook of the INS, 1992, 1993). Even if we consider the potential for delay because of INS backlogs, IRCA residents only accounted for 2.5 percent of immigrants who naturalized in FY The proportion of yearly naturalizations accounted for by IRCA immigrants did grow in 1995 and 1996, respectively, and peaked in 1997 when 23 percent of new naturalizations were IRCA related. Surprisingly, although IRCA residents represented more than 40 percent of all immigrants in fiscal years , they never accounted for more than 23 percent of naturalizations in any given year in the 1990s (Rytina, 2002). 9 The fact that Spanish-speaking Latin America has been the leading sending region of the world for unauthorized immigration to the United States and that California has been the leading state of residence of these immigrants resulted in significant Latino and California share of both IRCA applications and number of persons granted permanent residence. California accounted for over half (53.5 percent) of all IRCA applications nationwide. Although we do not have access to the number of total applicants from Latin America, we do know that they constituted 95 percent of the applicants from California (Legalization Application Processing System (LAPS), Statistics Division, INS). Nationally, Latinos comprised 88.3 percent of successful applicants (Immigration and Naturalization Service, The Triennial Comprehensive Report on Immigration, 1999; Statistical Yearbooks of the INS, ). The state-level geographic distribution of persons granted legal residency through IRCA is presented in Table In 1989, 1990, and 1991 IRCA LPRs were 44 percent, 57 percent, and 62 percent of all legal immigrants, respectively. 10 There is no published source with the IRCA LPR distribution by state after 1994, so the cumulative numbers presented are based on our estimates using other immigration statistics from the INS. The percent distribution does not change much there were only 13,097 persons granted residency through IRCA provisions between FY 1995 and FY 2001.

8 Effect of IRCA Citizens on Latino Voting 799 TABLE 1 Successful IRCA Applicants by State of Intended Residence n State of Intended Residence Percent of Nationwide California 53.27% Texas 15.09% New York 5.44% Florida 5.30% Illinois 5.35% New Jersey 1.49% Arizona 2.43% Other states 11.63% n Includes special agricultural workers (SAWs) and Pre-1982s. SOURCE: Immigration and Naturalization Service, The Triennial Comprehensive Report on Immigration, May 1999; Immigration and Naturalization Service, Statistical Yearbook of the INS, There was a tremendous increase in naturalizations in California and the United States during the 1990s. INS efforts to clear the backlog of applications as well as to speed up the naturalization process contributed significantly to the pattern that made 1996 the year of peak number of newly naturalized citizens. However, this does not fully explain the shift in the Latino percent of California naturalizations from 9.19 percent in 1990 to 50 percent in Prior to 1994, Latino immigrants in California constituted a smaller percent of those naturalizing than the national average. After 1994, Latinos in California constitute a much larger percent of new naturalized citizens. That being said, we used all available data to determine the number of IRCA LPRs of Latin American origin who successfully naturalized in the United States and estimates for this population among those naturalizing in California. Approximately 520,000 of the naturalized citizens between 1995 and 2001 were Mexican-born IRCA LPRs, with an estimated 300,000 residing in California (Rytina, 2002; U.S. Immigration and Naturalization Service, 2003). 11 Even if all new IRCA citizens registered to vote (which is highly unlikely; a better estimate might be about half ), they would only account for 30 percent of the estimated 1 million new Latino registered voters in California. The foregoing discussion suggests two possibilities concerning the IRCArelated naturalization phenomenon and Latino voting. On the one hand, IRCA may have the effect of increasing the pool of eligible Latino voters, which in turn leads to some amount of increase in the number of Latinos turning out to vote. On the other hand, IRCA may have a smaller effect than many anticipate, given the relatively small number of voters that could 11 The Statistical Yearbooks of the INS only gave a Mexican origin number of IRCA LPRs who naturalized.

9 800 Social Science Quarterly have been added through this process. It is also an open question whether Latinos naturalized through the IRCA process though newly eligible to vote constitute a pool of likely registrants and voters. That is, even were IRCA to increase the pool of eligible voters, it is not clear that these particular citizens are behind the recorded growth in the Latino electorate throughout the 1990s. As DeSipio (1996) notes, new citizens are not always civically minded. Our effort is intended to discern which of these possibilities finds more empirical purchase in the data. As we discussed at some length above, several factors may be, at least in part, related to the growth in Latino votes during the 1990s. The IRCA-based possibility as the leading cause for the increase in citizen population that drove the increase in the Latino vote has never been empirically examined. In testing the effect of IRCA on Latino voting, we hope to clarify the extent to which IRCA may be cited as a causal factor related to Latino voting. In the next section we discuss our data and approach to ascertaining the relationship between IRCA naturalizations and Latino voting. We move then to a presentation of our results and some discussion of what conclusions we can draw surrounding the IRCA effect. Data and Methodology The data for this project come from a combination of three traditional and unique sources. Numbers reported for registration and voter turnout are from the County Registrar of Voters in five southern California counties. 12 Population and demographic control variables are from Summary File 3 (SF-3) of the 2000 Census. Finally, data for our key independent variable, IRCA, come from Immigration and Naturalization Services (INS) unpublished statistics on IRCA petitions by country of origin, reported at the zipcode level. Because IRCA figures are available only by zipcode we aggregate both the County Registrar and Census data to the zipcode level and merge the three data sets into one file. In all, there are 496 zipcodes in the five-county area of study. Because the Registrar of Voters data and IRCA information are real data, as opposed to samples drawn from surveys, we have greater confidence in the relationships we report. We present two sets of OLS regression models for each of three dependent variables to determine the effect of IRCA on Latino vote growth. The dependent variables examined are (1) growth in the Latino vote ; (2) the Latino vote 1996; and (3) the Latino vote 2000, each of which is 12 The five counties are Los Angeles, Orange, San Bernardino, Riverside, and Ventura and represent 996,002 Latino voters, or 67 percent of all Latino voters in the State of California as of the 2000 election.

10 Effect of IRCA Citizens on Latino Voting 801 based on official vote results. Latino voters are differentiated from non- Latino voters by way of the Census Bureau Spanish Surname Database. 13 For each of the three dependent variables, two estimates are presented: a basic model testing the effect of IRCA in isolation against other demographic variables, and an expanded model with four additional controls, including an additional date of registration variable to account for the highly political context of the mid-1990s in California. All variables included in the model reflect total counts (as opposed to percentages) at the zipcode level. 14 We concentrate on total growth in the Latino vote because we are interested in substantive increases in Latino voting that percentage increases may mask. For example, if a zipcode reports 50 Latino voters in 1996 and 250 in 2000, this growth of 200 Latino voters is a 400 percent increase. At the same time, if a different zipcode reports 5,000 Latino voters in 1996 and 7,500 in 2000, this growth of 2,500 voters is a substantively more significant increase, but as a percentage it only reflects a 50 percent increase. Further, the conventional wisdom that the new IRCA citizens were driving the growth in Latino voting is based on the assumption that areas with big increases in Latino voters must be areas with large numbers of IRCA citizens. To test this hypothesis, the dependent variable (Latino vote) and the independent variable (IRCA petitions) reflect raw numbers. To account for two well-known problems in dealing with raw data multicollinearity and skewed distributions we transform most variables into their natural log (2.718) and employ log-log regression, consistent with Campbell (2000; see also Godfrey, McAleer, and McKenzie, 1988; Godfrey and Wickens, 1981). Although the logarithmic models are technically accurate, their coefficients are difficult to interpret with regard to the substantive effects of each independent variable. To address this concern, we retransform the beta coefficients by their natural log (ln beta ) to generate meaningful data for each variable. We explore this in more detail below in our discussion of the results. The first dependent variable tested, Vote Growth, is the sum of the total number of Latino voters in 2000 minus the total number in 1996 for each zipcode. The other two dependent variables are simply the total Latino vote counts for each year. To control for the relative size of the zipcode, two 13 The Spanish surname list is based on the 1990 Census and is constructed by tabulating the responses to the Hispanic origin question. Each surname is categorized by the percent of individuals that identified themselves as Hispanic. Each surname is then given a numeric value for the probability that persons with the surname are Hispanic. The list contains over 25,000 surnames. For a full explanation on the methodology of the list, see Word and Perkins (1996). 14 Raw numbers are used even for demographic controls such as education and age to keep all variables in the model consistent with one another. Given that the dependent variables, and key independent variables, are based on absolute numbers of Latinos within a zipcode, it is important that the related control variables are consistent.

11 802 Social Science Quarterly FIGURE 1 Relationship Between IRCA Applications (Log) and Latino Vote Growth (Log) 8 Latino Vote Growth Latino IRCA Applications variables, Total Population and Latino Population, are included, based on Census 2000 SF-3 data. As explained above, the key independent variable, IRCA Applications, is the raw number of petitions filed in accordance with IRCA by immigrants from Latin America within a given zipcode. Given that IRCA citizens report a higher degree of residential stability than non-irca foreign born (Rytina, 2002), we are not concerned that our results are being confounded by IRCA recipients who naturalized in one state and moved to another, where they became registered and voted. To specifically isolate the effect of IRCA voters on the overall growth of the Latino vote, it is essential to control for the Latino foreign-born and naturalized citizen population. Because there are likely to be large numbers of IRCA petitioners in zipcodes with large foreign-born and naturalized citizen populations, without controlling for naturalized citizens, the variable IRCA Applications becomes a proxy measure for all foreign-born voters. However, due to the high colinearity 15 between IRCA Applications, Naturalized Latinos, and Foreign-Born Latinos, including all three variables in the model will skew the individual contribution of each variable, making it impossible to determine the specific impact of IRCA. Thus, the log-trans- 15 In fact, the variables are correlated at po0.000 and when both are included in the model as continuous (unlogged) variables the mean variance inflation factor (VIF) for the model is 10.6, above the acceptable ceiling of 9.0. When all variables are logged, the mean VIF drops to 7.6, suggesting that colinearity is not a problem.

12 Effect of IRCA Citizens on Latino Voting 803 formation of all variables in the model is essential to alleviate concerns of collinearity, and also to keep the relationship between all variables consistent. For example, once the variables are transformed into their natural log, it appears that a linear relationship exists between IRCA applications and new Latino voters (see Figure 1). However, because zipcodes with high levels of IRCA applicants are also likely to have high levels of foreign-born and naturalized Latinos who did not gain citizenship through IRCA, it is important to include these additional control variables to determine the precise impact of IRCA vis-à-vis other cohorts of foreign-born Latino voters. Although the scatterplot shows a bivariate correlation between IRCA and Latino voting, this relationship needs to be more rigorously tested in a multivariate setting, as we do below. Beyond controlling for the 2000 population for each zipcode, it is necessary to control for the population growth, since we are estimating the growth in Latino voting. To this end, we include three growth measures based on the change in population from 1990 to 2000 for each zipcode. These variables include Total Pop. Growth, Latino Pop. Growth, and Foreign Pop. Growth. We include a number of other socioeconomic and demographic measures as controls. Median Latino Income reports the median income of Latino households within the zipcode as of Latino College is the total number of Latinos with an associates degree or higher and Age is the total number of registered voters over the age of 50 within the zipcode. Home Ownership is simply the total number of homeowners in a zipcode and Median Home Value is the average appraised value of all homes in the zipcode in Although there is still some collinearity in the models, it is not so high that it casts the results in doubt (see note 15). Our final two controls are intended to isolate the effect of two additional potential causes of increased voter turnout. Immig. Year of Entry captures the cohort of immigrants that had the opportunity to naturalize and vote by the 2000 election, but who were not eligible under IRCA provisions. These non-irca immigrant voters represent a competing hypothesis that it was not IRCA, but interest on the part of new citizens generally (perhaps in reaction to Proposition 187, see Segura, Falcon, and Pachon, 1997; Pantoja, Ramírez, and Segura, 2001; Barreto and Muñoz, 2003) that contributed to the growth in Latino voting in California. Latino Registrants measures newly registered Latino voters, those who registered in 1994, 1995, and 1996 (see Ramirez, 2002). Although newly registered voters are generally less likely to turn out, due to the heightened political environment for Latinos surrounding the 1994 reelection of Republican Governor Pete Wilson, the 1994 anti-immigrant ballot measure Proposition 187, and the 1996 anti-affirmative action ballot measure Proposition 209, we expect those Latinos that registered to vote during this time period to be more politically active (Pachon, 1998; Arteaga, 2000). Although some of the registration cohort could be comprised of IRCA recipients, most did not gain legal status and register to vote until

13 804 Social Science Quarterly after 1996, one of the primary reasons scholars have suspected IRCA citizens of contributing to the growth in the Latino electorate in the late 1990s. 16 Results and Analysis Table 2 displays the results from our regression models. In all estimates we use two-tailed tests of significance in estimating the relationships between the variables. For each model we report the coefficients and standard errors with standard notation for levels of significance. In addition, we report the transformations of the logged measures and the percentage change in Latino voting that results from a 100 percent increase in the independent variable in question (i.e., doubling the value of the independent variable, ceteris paribus, obtains a given percentage change in the number of Latino voters turning out in a zipcode). The percentages make possible a straightforward assessment of the substantive effect of each independent variable. We begin with our central question whether IRCA applications are associated with increases in Latino voter turnout. In each of the three estimates, the number of IRCA applicants produces no significant influence on voter turnout. Indeed, none of the models shows a result significant at even one standard deviation. The calculation of log-transformations bears out this lack of significant effect as well even doubling the number of IRCA applicants in a zipcode only produces between a 0.1 percent and 1.3 percent change in voter turnout. Substantively, all three results show that while IRCA applications may add to the citizen-eligible pool of Latinos, there is no evidence to suggest that these applicants go on to register or vote. Recall that we included two measures to proxy the influence of two alternative possibilities that interest on the part of new citizens generally (but not IRCA specifically) or the heightened political environment for Latinos resulting from contentious initiatives might be behind the increase in Latino votes being cast. Immig. Year of Entry shows an insignificant and negative result in the vote growth estimate, showing that the alternative immigrant cohort is not associated with change in Latino voting between 1996 and However, in each of the year-specific estimates, the 16 Previous research has suggested that the registration cohort is an important one to isolate because it came into the political system at a time of contentious politics for the Latino community (see Segura, Falcon, and Pachon, 1997; Tolbert and Hero, 2001; Pantoja, Ramirez, and Segura, 2001; Pachon, 1998; Barreto and Woods, 2005; Ramírez, 2002). It is necessary that we include this cohort of potential voters to fully test our hypothesis. Although there is some concern about overlap among IRCA citizens, most of the IRCA recipients did not come into the electorate until after According to Census Bureau estimates, twothirds of IRCA recipients did not achieve citizenship status until 1997 (see Rytina, 2002). Further, even with some overlap, our inclusion of this variable in the regression takes care of this by controlling out the effects of each group, given the contribution of each independent variable and holding all others at their mean.

14 Effect of IRCA Citizens on Latino Voting 805 TABLE 2 Influence of IRCA Naturalizations on Latino Vote Model 1 Model 2 Vote Growth Turnout 1996 Coef. Log-Trfm. Coef. Log-Trfm. Total population (log) nn nnn (0.0904) 26.0% (0.0519) 27.3% Latino population (log) nnn (0.0859) 54.1% (0.0494) 3.6% IRCA applications (log) (0.0258) 1.3% (0.0148) 1.3% Naturalized Latino (log) (0.0736) 5.9% (0.0423) 4.2% Foreign-born Latino (log) nn n (0.0956) 28.0% (0.0550) 12.5% Total pop. growth (log) n nnn (0.1262) 27.4% (0.0725) 23.8% Latino pop. growth (log) n (0.0889) 12.6% (0.0511) 11.7% Foreign pop. growth (log) n nnn (0.1410) 29.4% (0.0810) 25.1% Immig. year of entry (log) nn (0.0453) 2.2% (0.0260) 6.5% Latino registrants (log) nnn nnn (0.0549) 70.0% (0.0316) 104.3% Home ownership (log) nn (0.0453) 11.2% (0.0260) 3.2% Latino college (log) nnn nnn (0.0468) 17.5% (0.0269) 17.1% Age (log) nnn nnn (0.0356) 19.7% (0.0205) 7.6% Median Latino income (log) nnn nnn (0.0179) 6.4% (0.0103) 6.0% Residential stability (log) nnn nnn (0.0968) 51.4% (0.0556) 32.7% Constant (0.7875) 51.1% (0.4525) 29.1% N F Adj. R nnn po0.001; nn po0.010; n po coefficient for this measure is both positive and significant, suggesting that non-irca immigrants were likely to vote in both elections. The associated effect of doubling this measure is a 6.5 percent and 8.6 percent increase in Latino votes in 1996 and 2000, respectively. In combination, these results

15 806 Social Science Quarterly TABLE 2 continued Model 3 Turnout 2000 Coef. Log-Trfm. Total population (log) nnn (0.0322) 33.1% Latino population (log) nnn (0.0306) 18.3% IRCA applications (log) (0.0092) 0.1% Naturalized Latino (log) (0.0263) 0.9% Foreign-born Latino (log) nnn (0.0341) 18.0% Total pop. growth (log) (0.0450) 4.7% Latino pop. growth (log) n (0.0317) 6.4% Foreign pop. growth (log) (0.0503) 5.3% Immig. year of entry (log) nnn (0.0161) 8.6% Latino registrants (log) nnn (0.0196) 90.0% Home ownership (log) (0.0161) 0.3% Latino college (log) nnn (0.0167) 6.1% Age (log) nnn (0.0127) 21.6% Median Latino income (log) nnn (0.0064) 3.7% Residential stability (log) nnn (0.0345) 19.6% Constant (0.2808) 31.5% N 496 F Adj. R nnn po0.001; nn po0.010; n po demonstrate that while members of this alternative immigrant cohort produced votes in both 1996 and 2000, they did not contribute in a significant way to the growth of Latino votes across zipcodes between 1996 and Looking to the second alternative possibility, Latino Registrants shows a positive and significant effect across all three estimates. Of

16 Effect of IRCA Citizens on Latino Voting 807 particular note is the size of the effect in all three cases. In the vote growth estimate, the calculation of substantive effect shows that doubling the number of Latino registrants between 1994 and 1996 produces a 70 percent increase in the growth in Latino voters in a zipcode. In 1996, doubling this number obtains better than a 100 percent increase in Latino voters, and in 2000 the same measure is associated with a 90 percent increase in Latino voters. These figures are fairly remarkable, particularly when viewed relative to those we have reported to this point. Of the three possibilities we have considered, IRCA applicants appear to have no substantive effect, the non- IRCA immigrant cohort has a mild and positive effect, and the Latino registrants measure has a quite large and positive effect. These results show in a convincing manner that the IRCA reform, by itself, was not ultimately responsible for producing additional Latino votes in 1996 or 2000, or the growth in Latino votes over that timespan. Latinos naturalized outside of IRCA contributed to some degree to growth in Latino voting. However, most of the increase in Latino voting is associated with the increase in Latinos who registered between 1994 and 1996 a particularly divisive time for Latinos in California. Our control measures perform largely as expected. The size of the Latino population contributed to growth in the Latino vote, and the number of Latino voters in 2000, but had no significant effect in The naturalized Latino population not captured in the above measures has no significant influence on voting and the size of the foreign-born population exerts a consistently negative and significant influence on both growth in and the number of Latino voters. Interestingly, growth in the Latino population shows a significantly negative effect in the year-specific estimates, demonstrating that simple growth in the Latino population does not directly translate into Latino votes. Growth in the foreign-born population showed mixed results, showing a positive effect in the vote growth estimate, a negative effect in the 1996 estimate, and no significant effect in In terms of the socioeconomic measures, each of the year-specific measures showed a positive and significant coefficient except Home Ownership in a zipcode, which was insignificantly related to the number of Latino voters. In the vote growth estimates, the number of homeowners and residents over age 50 contributed to growth in the Latino vote, while the Latino College, Median Latino Income, and Residential Stability measures produced negative and significant effects on growth. These last measures show that where residential stability and Latino incomes are high, or Latino college graduates are numerous, there is already a relatively large number of Latinos voting and little room for upward growth. Recall that there are four general explanations offered for the increase in Latino voting through the 1990s: (1) Latino population growth, (2) citizens newly naturalized, particularly through the IRCA process, (3) mobilization efforts, and (4) the political context. Our results suggest that of these (ex-

17 808 Social Science Quarterly cluding mobilization efforts, which are not explicitly included in the model), the IRCA-based explanation is least compelling. In fact, our results show that IRCA applicants are not associated with increases in Latino votes at all. Our results suggest a different result for the later cohort of immigrants those who immigrated between 1985 and 1994 which are associated with modest increases in Latino votes in the year-specific estimates. These two results together indicate that immigrants who actively sought out U.S. citizenship without the benefit of IRCA, in contrast to immigrants who were encouraged to naturalize under the program, were a contributing factor to increasing the Latino vote. Put another way, newly naturalized Latinos helped increase the Latino vote, but not those who originally obtained legal resident status through IRCA provisions. These estimates suggest that the political context present in California had an effect on increasing Latino votes. The variable counting Latino registrants between 1994 and 1996 is not only significant, but shows a far more robust result than our other variables of central interest. This result is consistent with previous research on Latino voters in California (Barreto and Woods, 2005; Ramirez, 2002) and confirms that the increase in Latino voting in the 1990s was not simply a product of an increased pool of citizen-eligible Latinos, but instead due to an increase in interest among Latino voters. Conclusions We began this article by charting the increases in Latino voting in California and reviewing some of the leading explanations offered for this growth. One of the most often-cited, yet underexamined, explanations points to the influence of the Immigration Reform and Control Act of Here, we find that to the extent that newly naturalized citizens help drive the increases in Latino voting, it is not the result of the IRCA amnesty process. In fact, increases in IRCA applications are actually not significantly related to voter turnout in 1996 and 2000, and the growth in the Latino vote between 1996 and However, our alternative immigrant cohort measure is positively related to increases, showing that newly naturalized citizens are behind some of the increasing Latino vote, but not the segment naturalized through the IRCA process. The single most important factor emerging from our estimates, however, is the measure accounting for newly registered Latinos in the 1994 through 1996 period. It appears that Latinos who came into the electorate at a time of contentious politics for the Latino community demonstrated a continued interest in voting and were a driving force behind the Latino vote growth in the 1990s. Several implications may be drawn from the current work, and several additional inquiries may be suggested. To begin with, our work makes clear

18 Effect of IRCA Citizens on Latino Voting 809 that the increases in Latino voting in California were not the result of the IRCA amnesty. This, from our standpoint, bodes well for the future of Latino political incorporation. If the documented increases were due in large part to the IRCA amnesty (a one-time policy), the increases may not be sustainable into the future. At the very least, any increases would rely on an active intervention of government, which may or may not happen again. If, instead, the Latino community is simply more active and politically involved than it was in the past, it is more likely that Latinos will continue to grow in terms of votes, and as a share of the electorate. Similarly, while the effect of the anti-latino political context present in the state probably drove much of the increase in Latino political participation, it is not clear yet whether the momentum spawned is sustainable into the future as well. The fact that new immigrants in the non-irca cohort were positively associated with vote increases is perhaps our most optimistic finding. We take this to indicate that non-irca legal immigrants who choose to naturalize, and actively seek to do so, are more likely to eventually register and vote than those who qualify for naturalization through an amnesty program. Although this article has examined the 1986 IRCA cohort of immigrants, it holds implications for the future, as the federal government continues to debate the best approach to incorporating the more than 7.5 million undocumented Latino immigrants living in the United States. When a new resolution is implemented, further research will be warranted to determine what impact future generations of legalized immigrants have on the Latino electorate. REFERENCES Arteaga, Luis To Vote or Not to Vote? An Examination of Latino Voting Patterns and Motivations to Vote. Civic Participation Report. San Francisco, CA: Latino Issues Forum. Barreto, Matt A., and Jose A. Muñoz Reexamining the Politics of In-Between : Political Participation Among Mexican Immigrants in the United States. Hispanic Journal of Behavioral Sciences 25: Barreto, Matt A., Gary M. Segura, and Nathan D. Woods The Mobilizing Effect of Majority-Minority Districts on Latino Turnout. American Political Science Review 98: Barreto, Matt A., and Nathan D. Woods The Anti-Latino Political Context and its Impact on GOP Detachment and Increasing Latino Voter Turnout in Los Angeles County. In Gary Segura and Shawn Bowler, eds., Diversity in Democracy: Minority Representation in the United States. Charlottesville, VA: University of Virginia Press. Campbell, Cameron Intermediate Statistical Methods: Interpreting Coefficients for Transformed Variables. Los Angeles, CA: University of California. Available at hwww.soc.ucla.edu/faculty/campbell/210b_winter2000/210b_winter2000_class_8.pdfi. de la Garza, Rodolfo, Charles Haynes, and Jason Ryu Patterns of Latino Turnout: Study of the General Elections Harris County, Texas. Claremont, CA: Tomás Rivera Policy Institute.